AustraliaCountry of destination:
The ecosystem services that an agricultural landscape can provide to human beings are various and sometimes antagonists (all of them cannot be optimized simultaneously): food provision, water quality, recreational services, patrimonial biodiversity. When managing agro-ecosystems, finding “good compromise” solutions, in terms of services is therefore critical. However, finding landscape management strategies that enable to reach satisfying trade-offs on different services is difficult for several reasons: (i) the processes involved in the provisioning of these services are spatio-temporal processes (ii) decision must been made under uncertainty (iii) modeling and solving multi-objective optimization problems is more difficult that mono-objective ones. The goal of the project is to develop a decision aiding approach that encompass all of these issues and that can be applied to the understanding of trade-offs between production, regulation and pollination services related to management strategies of land use (crop field, grassland, semi-natural habitat) in a farming landscape. These objectives require methodological breakthrough with multi-objective optimization and performance guarantees about the quality of the solutions. Another challenge is to model a case study involving ecologists from the DYNAFOR unit, INRA Toulouse. The problem is to find the “best” dynamic management of crops considering the influence of some ecosystems as pollinators, their dynamics, and the influence of crops on these ecosystems. Allowing the indirect control of pollinators could for example reduce the need of pesticides. Because there are several criteria there is not only one best possible management, and a tool allowing interactions with decision-makers seems relevant to develop.
I received my PhD in 2013 at Paris Dauphine University in computer sciences notably about multi-objective and multi- mode traffic control on single junctions, being also in the same time assistant lecturer in mathematics. Then I got a position as post doc fellows up to 2017, at the CSIRO (Dutton Park, Australia) in the Iadine Chades’s research group in computer science and ecology. My main research themes are planning under uncertainty, combinatorial optimization and multi- objective optimization. I try to combine these approaches as much as possible when modelling and solving complex decision problems in ecology and transportation. During my AgreenSkills postdoc, I will use my previous experience to model and solve a complex decision problem of agro-ecology containing all aspects I previously studied. Solving exact spatial multi-objective planning problems under uncertainty was never really tackled in agro-ecology and it would be a natural continuation of my previous works.
Yann Dujardin, Iadine Chadès, 2018. Solving multiobjective optimization problems in conservation with the reference point method. PLoS ONE 13(1): e0190748. Doi: 10.1371/journal.pone.0190748.
Yann Dujardin, Tom Dietterich, Iadine Chadès, 2017. Three new algorithms to solve N-POMDPs. Proceedings of the 31th Association for the Advancement of Artificial Intelligence conference (AAAI), 4495-4501, AAAI Press.
Iadine Chadès, Sam Nicol, Tracy M. Rout, Martin Péron, Yann Dujardin, Jean-Baptiste Pichancourt, Alan Hastings, and Cindy E. Hauser, 2016. Optimization methods to solve adaptive management problems. Theoretical Ecology, 1-20. Doi: 10.1007/s12080-0160313-0.
Hawthorne Beyer, Yann Dujardin, Matthew Watts, Hugh Possingham, 2016. Solving conservation planning problems with integer linear programming. Ecological modelling, 328:14-22.
Yann Dujardin, Tom Dietterich, and Iadine Chadès, 2015. Alpha-min: a compact approximate solver for finitehorizon POMDPs. Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI), 25822588, AAAI Press.